The ability to accurately classify brain tumors in medical scans is crucial. It allows for faster and more precise detection of cancerous tissue. This research explores using a deep learning technique called VGG16 to categorize brain tumors. We made use of an MRI imaging collection that included four different forms of brain tumors: Giloma, Pituitary, Meningioma, and healthy brain tissue. By fine-tuning the VGG16 model, we achieved an impressive accuracy of 99.1%, demonstrating its effectiveness in brain tumor classification. However, to make this deep learning model more useful in clinical settings, we need to understand how it arrives at its decisions. To address this, we implemented a technique called Layer-wise Relevance Propagation (LRP). LRP helps visualize the specific areas in the brain scans that most influence the model’s predictions. This transparency is essential for doctors, as it allows them to see the reasoning behind the model’s diagnosis and builds trust in these automated tools. In conclusion, our research shows that combining VGG16 with LRP offers a powerful solution. This approach maintains high classification accuracy while also providing interpretable results. This highlights the potential of combining advanced deep learning models with explainability techniques to improve both accuracy and transparency in medical image analysis.